Expression Profiling and Differential Screening Between

Oncogene (2004) 23, 5901–5911 & 2004 Nature Publishing Group All rights reserved 0950-9232/04 $30.00 www.nature.com/onc

Expression profiling and differential screening between hepatoblastomas and the corresponding normal livers: identification of high expression of the PLK1 oncogene as a poor-prognostic indicator of hepatoblastomas

Shin-ichi Yamada1, Miki Ohira1, Hiroshi Horie2, Kiyohiro Ando1, Hajime Takayasu1, Yutaka Suzuki3, Sumio Sugano3, Takahiro Hirata4, Takeshi Goto4, Tadashi Matsunaga2, Eiso Hiyama2, Yutaka Hayashi2, Hisami Ando2, Sachiyo Suita2, Michio Kaneko2, Fumiaki Sasaki2, Kohei Hashizume2, Naomi Ohnuma2 and Akira Nakagawara*,1,2

1Division of Biochemistry, Chiba Cancer Center Research Institute, Chiba 260-8717, Japan; 2Japanese Study Group for Pediatric Liver Tumor, Japan; 3Human Genome Center, Institute of Medical Science, University of Tokyo, Tokyo 108-8639, Japan; 4Hisamitsu Pharmaceutical Co. Inc., Tokyo 100-6221, Japan

Hepatoblastoma is one of the most common malignant the prognosis of hepatoblastoma. Thus, the differentially liver tumors in young children. Recent evidences have expressed genes we have identified may become a useful suggested that the abnormalities in Wnt signaling path- tool to develop new diagnostic as well as therapeutic way, as seen in frequent mutation of the b-catenin gene, strategies of hepatoblastoma. may play a role in the genesis of hepatoblastoma. Oncogene (2004) 23, 5901–5911. doi:10.1038/sj.onc.1207782 However, the precise mechanism to cause the tumor has Published online 28 June 2004 been elusive. To identify novel hepatoblastoma-related genes for unveiling the molecular mechanism of the Keywords: hepatoblastoma; expression profile; oligo- tumorigenesis, a large-scale cloning of cDNAs and capping cDNA library; PLK1; prognostic factor differential screening of their expression between hepato- blastomas and the corresponding normal livers were performed. We constructed four full-length-enriched cDNA libraries using an oligo-capping method from the Introduction primary tissues which included two hepatoblastomas with high levels of alpha-fetoprotein (AFP), a hepatoblastoma Hepatoblastoma (HBL) is the most common hepatic without production of AFP, and a normal liver tissue cancer in children (Exelby et al., 1975; Weinberg and corresponded to the tumor. Among the 10 431 cDNAs Finegold, 1983). However, the etiology of HBL has been randomly picked up and successfully sequenced, 847 unclear in contrast to the adult hepatocellular carcino- (8.1%) were the genes with unknown function. Of interest, ma (HCC), in which preceding infection of hepatitis the expression profile among the two subsets of hepato- virus is often found (Buendia, 1992; Idilman et al., blastoma and a normal liver was extremely different. A 1998). Although most HBLs are sporadic, it is some- semiquantitative RT–PCR analysis showed that 86 out of times associated with certain hereditary diseases such as 1188 genes tested were differentially expressed between Beckwith–Wiedemann syndrome (Albrecht et al., 1994) hepatoblastomas and the corresponding normal livers, but and familial adenomatous polyposis (Li et al., 1987; that only 11 of those were expressed at high levels in the Giardiello et al., 1996; Kinzler and Vogelstein, 1996). In tumors. Notably, PLK1 oncogene was expressed at very the former, loss of heterozygosity of chromosome high levels in hepatoblastomas as compared to the normal 11p15.5 is frequently observed, and the abnormal infant’s livers. Quantitative real-time RT–PCR analysis regulation of the insulin-like growth factor 2 (IGF2) for the PLK1 mRNA levels in 74 primary hepatoblasto- and the H19 genes at this locus may contribute to the mas and 29 corresponding nontumorous livers indicated disease (Albrecht et al., 1994; Montagna et al., 1994; Li that the patients with hepatoblastoma with high expres- et al., 1995; Rainier et al., 1995; Yun et al., 1998; sion of PLK1 represented significantly poorer outcome Fukuzawa et al., 1999). In the latter, the APC gene, than those with its low expression (5-year survival rate: which is one of the key molecules in Wnt signaling, was 55.9 vs 87.0%, respectively, p ¼ 0.042), suggesting that found to be constitutively mutated (Kinzler and the level of PLK1 expression is a novel marker to predict Vogelstein, 1996). Increasing evidence suggests that Wnt signaling pathway also plays an important role in the genesis of *Correspondence: A Nakagawara, Division of Biochemistry, Chiba sporadic hepatoblastomas. A high frequency (more than Cancer Center Research Institute, 666-2 Nitona, Chiba 260-8717, Japan; E-mail: [email protected] 60% in some reports) of somatic mutations in the b- Received 9 December 2003; revised 26 March 2004; accepted 1 April catenin gene has recently been reported in sporadic 2004; published online 28 June 2004 tumors (Koch et al., 1999; Wei et al., 2000; Takayasu Differential screening of hepatoblastoma cDNA libraries S Yamada et al 5902 et al., 2001; Buendia, 2002). Mutant b-catenin proteins using full-length-enriched oligo-capping cDNA li- accumulate in the nucleus, resulting in stimulating braries. In the present study, we have identified 86 transcription of the target genes such as c-myc and genes differentially expressed between HBLs and their cyclin D1 (Morin et al., 1997; Polakis, 1999). Mutation corresponding normal livers. One of such genes, PLK1, in the Axin gene, whose product is an antagonist of showed a significantly high expression in the formers as nuclear accumulation of b-catenin, has also been found compared with the latters, and its high expression was in HBL and may contribute to the pathogenesis of the significantly associated with poor prognosis of HBLs. tumors without b-catenin mutation (Taniguchi et al., 2002; Miao et al., 2003). However, the molecular mechanism underlying the pathogenesis of HBL is still Results largely unknown. Recent progress in therapeutic strategies including Expression profiles of primary HBLs and a normal liver intensive chemotherapy and liver transplantation im- proved the outcome of the patients with HBL. However, To obtain the genes expressed in primary HBLs and the prognosis of a significant fraction of the tumors still normal infant’s liver, we constructed oligo-capping remains poor. The clinical markers currently used for cDNA libraries from two primary HBLs with increased HBL include staging, which is a major instrument for AFP secretion (HMFT, HYST), a primary HBL with- assessing prognosis (Hata, 1990), serum alpha-fetopro- out AFP secretion (HKMT), and a corresponding tein (AFP) (Mann et al., 1978), mitotic activity (Haas normal liver (HMFN). After cloning 3000 cDNAs from et al., 1989), DNA ploidy (Hata et al., 1991), nuclear each of the four cDNA libraries, 2289, 2837, 2537, and localization of b-catenin (Park et al., 2001), p53 2768 clones from the libraries of HMFT, HYST, mutation (Oda et al., 1995), and chromosomal altera- HKMT, and HMFN, respectively, were successfully tion (Weber et al., 2000). Serum AFP level is used as a end-sequenced. Homology search against the public diagnostic marker to monitor the tumor progression, databases of those 10 431 clones by BLAST program responsiveness to the therapy, and recurrence after the revealed that 847 clones (8.1%) in total contained novel treatment. Extremely high levels of serum AFP are sequences which had not been annotated (Table 1). reported to be associated with aggressiveness of the To elucidate the gene expression pattern in each tumors with unfavorable outcome (van Tornout et al., cDNA library, we compared expression profile of the 1997), except some reports showing that there is no known genes that appeared in three different kinds of significant relationship between initial serum AFP levels libraries, a HBL with positive AFP (HMFT), a HBL and prognosis of the patients with HBL (Ortega et al., with negative AFP (HKMT), and an infant’s liver 1991; von Schweinitz et al., 1994). Moreover, the tumor (HMFN) (Table 2). BodyMap (Okubo et al., 1992) and with low levels of serum AFP often grows rapidly and is a serial analysis of gene expression (SAGE) (Velculescu often reluctant to chemotherapy (von Schweinitz et al., et al., 2000) are very good methods to quickly provide 1995). The other genetic markers including DNA ploidy, quantification of the levels of all mRNAs in certain chromosomal aberration, and p53 mutation are not so tissues and cell types by high throughput end-sequencing powerful clinical indicators. Even the nuclear localiza- of cDNA clones. In this study, we applied the former tion of b-catenin and/or mutation of the b-catenin gene method by counting cDNA clones to show each appear to lose their impact as a prognostic factor when expression profile of HBL tumors or a non-tumorous combined with the grade of histological differentiation tissue. Although each library consists of 3000 clones, because of its close correlation with the latter (Takayasu which may be a rather small number, the frequency of et al., 2001). Therefore, we may need to find novel each cDNA appearance provides a hint to understand markers to predict the patient’s outcome in a compre- each tissue’s genetic background. hensive way. Overall, the most frequently appeared gene was To understand the molecular mechanism of the albumin as expected, which was extremely low in the genesis and progression of HBL, as well as to develop tumor with negative AFP. Genes involved in cellular a novel diagnostic and therapeutic system for the tumor, structure and/or maintenance, glucose and lipid meta- we have randomly cloned 10 431 cDNAs expressed in bolisms, and a part of protein synthesis and its transport primary HBL tissues and a normal infant’s liver by were frequently found in the normal liver library. On the

Table 1 Summary of the number of genes cloned from the cDNA libraries of hepatoblastomas and a normal infant liver of hepatoblastomas Oligo-capping cDNA library No. of the clones No. of the genes No. of the successfully genes with end-sequenced unknown function

Hepatoblastomas with positive AFP 6000 5126 323 (6.3%) Hepatoblastoma with negative AFP 3000 2537 262 (10.3%) Infant’s liver 3000 2768 262 (9.5%)

Total 12 000 10 431 847 (8.1%)

Oncogene Differential screening of hepatoblastoma cDNA libraries S Yamada et al 5903 Table 2 Comparison of the known genes frequently appeared in hepatoblastomas with or without secretion of AFP and a non-tumorous infant’s liver No. of appearance of the genes

Gene symbol Acc. no. Gene name HBL with Normal HBL with positive infant’s negative AFP liver AFP

Total number of genes 2289 2768 2537

Protein synthesis, metabolism, transport ALB NM_000477 Albumin 558 482 8 AFP NM_001134Alpha-feto protein 67 0 0 AGT NM_000029 Angiotensinogen 43 16 0 EEF1A1 X03558 Eukaryotic translation elongation factor 1 alpha 1 35 20 87 RPL27A NM_000990 60S ribosomal protein L27a 31 452 FTL M11147 Ferritin 24 11 3 FGA NM_021871 Fibrinogen, A alpha polypeptide 20 38 2 HP K01763 Haptoglobin 19 6 1 ORM1 X02544 Orosomucoid-1 12 8 0 RPS27 NM_001030 Ribosomal protein S27 11 431 F2 J00307 Coagulation factor 2 11 26 0 TF NM_001063 Transferrin 8 6 0 PAH U49897 Phenylalanine hydroxylase 6 6 0 PLG NM_000301 Plasminogen 5 8 0 SERPINA1 X01683 Serine proteinase inhibitor, clade A, member 1 5 6 0 GC NM_000583 Group-specific component 421 1 RPS29 NM_001032 Ribosomal protein S29 3 1 0 CTSB NM_147783 Cathepsin B 2 5 3 SERPING1 BC011171 Serine proteinase inhibitor, clade G, menber 1 2 33 0 CRP X56692 C-reactive protein 1 8 0 ITIH2 NM_002216 Inter-alpha (globulin) inhibitor, H2 polypeptide 0 25 0

Growth factor MST1 M74178 Macrophage stimulating 1 8 16 0

Cell signaling WIF1 NM_007191 Wnt inhibitory factor 1 0 0 11 DKK1 NM_012242 Dickkopf 0 0 7

Cell structure, adhesion VTN NM_000638 Vitronectin 7 30 0 ACTB BC013380 Actin 6 17 6 LRG AF403428 Leucine-rich alpha-2-glycoprotein 6 11 0 VIM NM_003380 Vimentin 0 3 38

Cell cycle RBM4NM_002896 RNA binding motif protein 2 0 21 RAP1B NM_015646 RAP1B 0 0 11

Organism defense BF L15702 B-factor, properdin 5 13 0 GPX1 NM_000581 Glutathione peroxidase 40 0 C1R NM_001733 Complement component 1 1 21 1

Glycometabolism LDHA NM_005566 Lactate dehydrogenase 19 28 7 ADH1B AF153821 Alcohol dehydrogenase 15 29 1 CES1 L07764Carboxylesterase 9 22 2 ALDH1A1 NM_000689 Aldehyde dehydrogenase 2 13 2

Lipid metabolism EPHX1 NM_000120 Epoxide hydrolase 1 7 12 0 APOA2 NM_001643 Apolipoprotein A-II 6 2 0 ADFP BC005127 Adipose differentiation-related protein 5 141

Heat shock protein, metabolic enzyme UGT2B4Y00317 UDP-glucuronosyltransferase 11 32 2 HSPA8 NM_006597 Heat shock 70 kDa protein 1 6 1

Unknown, others ATP5A1 NM_004046 ATP synthase 18 11 23 SEPP1 NM_005410 Selenoprotein P 7 10 2

Oncogene Differential screening of hepatoblastoma cDNA libraries S Yamada et al 5904 Table 2 (continued ) No. of appearance of the genes

Gene symbol Acc. no. Gene name HBL with Normal HBL with positive infant’s negative AFP liver AFP

CYP3A4M18907 P450 6 81 3 AHSG M16961 Alpha-2-HS-glycoprotein 6 5 2 TPT1 X16064Translationally controlled tumor protein 6 0 3 CYP2C9 M61855 P4502C9 1 10 1

other hand, genes involved in protein synthesis such as elongation factors and ribosomal proteins were ob- served more frequently in HBLs than in normal liver. The expression profile in the library of the tumor without AFP secretion was very different from that with positive AFP (HMFT vs HKMT). As expected, AFP gene did not appear in the HKMT library. Intriguingly, Wnt Inhibitory factor-1 and dickkopf, both of which are inhibitors of Wnt signaling (Hsieh et al., 1999; Wang et al., 2000), frequently appeared in the HKMT library. In addition, vimentin, RNA-binding motif protein,and RAP1B also frequently appeared in the HKMT library, but hardly in the HMFT library with AFP secretion. Thus, HBL with positive AFP and that with negative AFP seem to have a distinct gene expression profile, resulting in different biological characteristics.

Identification of the differentially expressed genes between HBLs and normal livers To identify differentially expressed genes between HBLs and their corresponding normal livers, 1188 independent genes which included all of the 847 genes with unknown function and 341 known genes that were related to cellular functions including cell growth and differentia- tion among the 10 431 cDNAs were selected and subjected to semiquantitative RT–PCR analysis (Figure 1a). The complementary DNAs reverse-tran- scribed from total RNA obtained from eight tumors and their corresponding normal livers were used as PCR Figure 1 Expression of the representative genes by semi-quanti- tative RT–PCR. (a) Differentially expressed genes between HBLs templates after normalization with GAPDH expression. with or without b-catenin mutation and the corresponding normal As a result, we found that 75 genes were expressed at livers. cDNA was synthesized from RNAs prepared from eight higher levels in normal livers than in HBLs, whereas pairs of tumors and their corresponding normal livers, and was only 11 genes were expressed at higher levels in the used as a PCR template. Amount of cDNAs was normalized to that of GAPDH. Four tumors (cases 14, 67, 78, and 81) were with tumors than in normal livers. Figure 1a shows the wild-type b-catenin gene, while the other four tumors (cases 10, 58, representatives of the results of differential screening 77, and 85) were with mutant b-catenin gene. Gene symbols were using semi-quantitative RT–PCR and Table 3 lists 46 shown on the left; CRP: C-reactive protein, ALDOB: aldolase, CP: differentially expressed genes with known functions. We ceruloplasmin, NPC1: Niemann–Pick disease, type C1, OLR1: classified those differentially expressed genes into 12 oxidized low-density lipoprotein receptor 1, LAK: lymphocyte alpha-kinase. N: normal, T: tumor. (b) Semiquantitative RT–PCR categories according to their known functions. The of multiple human tissues. HMFT0601 exhibited ubiquitous genes preferentially expressed in normal liver showed expression in all tissues examined, whereas HMFN1655 and the profiles which reflected normal liver function. HMFT1272 showed specific expression in liver and fetal liver Consistent with the previous reports about HBL and hepatocellular carcinoma (von Horn et al., 2001; Xu et al., 2001; Kinoshita and Miyata, 2002), Insulin-like activity, is upregulated in HBLs, suggesting that the growth factor binding protein-3 (IGFBP-3), aldolase B, IGF axis may be involved in development of the tumor ceruloplasmin, and c-reactive protein were downregu- (Gray et al., 2000). lated in HBLs as compared with the normal livers. The Four known genes which were expressed at high levels expression of IGF2, whose product has mitogenic in HBLs (tumor4normal liver) include GTP-binding

Oncogene Differential screening of hepatoblastoma cDNA libraries S Yamada et al 5905 livers, we further examined the role of its expression in HBL.

PLK1 oncogene is overexpressed in HBLs Recent studies have demonstrated that the preferential expression of PLK1 mRNA is associated with some cancers including non-small-cell lung cancer (Wolf et al., 1997), squamous cell carcinoma of the head and neck (Knecht et al., 1999), and esophageal carcinoma (Tokumitsu et al., 1999). However, the role of PLK1 in HBL has never been reported. As indicated by semi- quantitative RT–PCR described above, we found that PLK1 mRNA expression in HBLs is higher than in normal livers (Figure 2a). Northern blot analysis also Figure 2 Increased expression of PLK1 in HBLs. (a) Semiquanti- confirmed its higher expression in HBLs (Figure 2b). We tative RT–PCR of PLK1 gene in eight HBL cases. Preferential also performed Southern blot analysis by using the expression of the PLK1 was seen in all sample pairs with and genomic DNAs obtained from primary HBLs and without b-catenin mutation. (b) Northern blot analysis of PLK1 in primary HBLs, The 28S ribosomal band is shown as a control of human placenta as a control, and probed with the each RNA amount PLK1-specific DNA fragment. However, we failed to find any clue of rearrangements or amplification of the PLK1 gene locus (data not shown). nuclear protein gene RAN, PLK1 onocgene, and two To examine the clinical significance of the expression cholesterol metabolism-associated protein genes, low- level of PLK1, we performed quantitative real-time RT– density lipoprotein (LDL) receptor 1 and Niemann-Pick PCR analysis using 74primary hepatoblastomas and 29 disease type C1 (NPC1). The RAN protein is involved in corresponding normal liver samples (Figure 3a). The the control of nucleo-cytoplasmic traffic of many average arbitrary values of PLK1 expression in HBLs nuclear proteins through formation of the transport and normal livers were 28.976.7 and 4.170.76, nuclear pore complex (Ribbeck et al., 1998). Nagata respectively (mean7s.e.m., Po0.01). The average va- et al. (2003) also reported that RAN is upregulated in lues in alive and dead cases were 21.775.2 (n ¼ 61) and HBLs by oligonucleotide DNA array experiment. The 62.4728.2 (n ¼ 13), respectively (p ¼ 0.021). When we LDL receptor 1 binds LDL, a major plasma cholesterol- compared the expression levels of PLK1 between 24- carrying lipoprotein, and plays an important role in paired HBLs and their corresponding normal livers, the cholesterol homeostasis (Sudhof et al., 1987; Goldstein former in HBL samples was significantly higher in and Brown, 1990; Hamanaka et al., 1992). NPC1 is a comparison with the latter (Po0.01) (Figure 3b). We causal gene of Niemann–Pick type C disease which is an also examined the relationship between the expression autosomal recessive lipid storage disorder that affects levels of PLK1 and clinicopathological data of HBLs. the viscera and central nervous system (Brady et al., Statistically significant correlation was observed only 1989). It encodes a protein with sequence similarity between histology and PLK1 expression (p ¼ 0.041). The to the morphogen receptor ‘patched’, and to the expression level of PLK1 in the tumors with poorly cholesterol-sensing regions of 3-hydroxy-3-methylglu- differentiated histology was higher than those with the taryl coenzyme A (HMG-CoA) reductase (Loftus et al., well-differentiated one. The other clinicopathological 1997) and is involved in the intracellular trafficking of factors such as age, clinical stage, and b-catenin cholesterol. Concerning the differentially expressed mutation did not show a statistical significance with genes which contained unknown sequences, those PLK1 expression. cDNA sequences have been submitted to the public To further examine whether the PLK1 expression was database (Genbank/DDBJ Accession numbers: AB07 associated with the outcome of the patients with HBL, 3346-AB073347, AB073382-AB073387, AB073599-AB0 we performed a Kaplan–Meier analysis (Figure 4). The 73614, and AB075869-AB075881). Interestingly, only distinction between high and low levels of PLK1 one known gene, lymphocyte alpha-kinase (LAK), expression was based on the median value (low, showed distinct expression pattern between HBLs with PLK1o13 d.u.; high, PLK1Z13 d.u.). Since the overall mutant b-catenin and those with wild type b-catenin survivals of 15 out of 74cases were unknown, 59 cases (Figure 1a). were applied to the analysis. The 5-year survival rates of We next examined expression pattern of the novel the groups with high and low PLK1 expression were genes in human multiple tissues by semi-quantitative 55.9 and 87.0%, respectively (P ¼ 0.042). The univariate RT–PCR and found that at least five genes were analysis showed that both PLK1 expression (P ¼ 0.015) specifically expressed in the liver (a part of the data is and histology (P ¼ 0.025) have a significant prognostic shown in Figure 1b). Since the oncogene PLK1 (polo- importance (Table 4). The multivariate analysis demon- like kinase-1) was expressed in HBLs at significantly strated that PLK1 expression was significantly related to high levels as compared with the corresponding normal survival, after controlling b-catenin mutation, age, stage,

Oncogene Differential screening of hepatoblastoma cDNA libraries S Yamada et al 5906 Table 3 The known genes differentially expressed between hepatoblastomas and normal livers Gene symbol Acc. no Gene name

Protein synthesis, metabolism, transport T4N RAN NM_006325 GTP-binding nuclear protein RAN N4T LBP AF105067 Lipopolysaccharide-binding protein N4T TDO2 BC005355 Tryptophan 2,3-dioxygenase N4T CRP X56692 C-reactive protein N4T GC NM_000583 Group-specific component N4T HP K01763 Haptoglobin N4T HPX NM_000613 Hemopexin N4T SQSTM1 NM_003900 Sequestosome 1 N4T PHDGH AF171237 A2-53-73 3-phosphoglycerate dehydrogenase N4T PPP1R3C XM_005398 Protein phosphatase 1, regulatory (inhibitor) subunit 3C N4T ITIH4D38595 Inter-alpha-trypsin inhibitor family heavy chain-related protein N4T G1P2 M13755 Interferon-induced 17-kDa/15-kDa protein Cytokine, growth factor, hormones N4T HABP2 D49742 Hyaluronan binding protein 2 N4T IGFBP3 NM_000598 Insulin-like growth factor binding protein 3 N4T GOT1 AF052153 Glutamic-oxaloacetic transaminase 1 Cell signaling N4T CSNK2B M30448 Casein kinase II, beta polypeptide N4T TPD52 NM_005079 Tumor protein D52

cell cycle T4N PLK1 X73458 PLK1

Cell structure, adhesion N4T LRG AF403428 Leucine-rich alpha-2-glycoprotein N4T PGRP-L AF384856 Peptidoglycan recognition protein L precursor N4T CLDN4NM_001305 Claudin4 N4T VTN NM_000638 Vitronection

Organism defense N4T RODH-4NM_003708 Retinol dehydrogenase 4 N4T MASP1 AF284421 Mannan-binding lectin serine protease 1 N4T C4BPA M31452 Complement component 4 binding protein, alpha

Glycometabolism N4T ADH1B AF153821 Alcohol dehydrogenase 1B, beta polypeptide N4T ALDOB M15657 Aldolase B

Lipid metabolism T4N NPC1 NM_000271 Niemann–Pick disease, type C1 T4N OLR1 NM_002543 Oxidized low density lipoprotein (lectin-like) receptor 1 N4T DGAT2 AF384161 Diacylglycerol acyltransferase N4T SCP2 NM_002979 Sterol carrier protein 2 N4T APOA5 AF202890 Apolipoprotein A-V N4T AADAC L32179 Arylacetamide deacetylase N4T SAA4M81349Amyloid A protein

Transcription N4T BZW1 NM_014670 Basic leucine zipper and W2 domains 1 N4T CREB-H NM_032607 CREB/ATF family transcription factor

RNA biogenesis, metabolism N4T HNRPDL AB017018 Heterogeneous nuclear ribonucleoprotein D-like

Homeostasis, heat shock protein, metabolic enzymes N4T UGT1A AF297093 UGT1 gene locus N4T ALPL X14174 Liver-type alkaline phosphatase N4T SLC10A1 L21893 Solute carrier family 10 N4T CES1 AF177775 Carboxylestelase N4T AKR1D1 Z28339 Aldo-keto reductase family 1, member D1 N4T AKR1C2 U05598 Aldo-keto reductase family 1, member C2 N4T CP D45045 Ceruloplasmin

Others N4T DGCR6L NM_033257 DiGeorge syndrome critical region gene 6 like N4T A1BG AF414429 Alpha-1-B glycoprotein

T4N: highly expressed in the tumors as compared to normal livers. N4T: highly expressed in normal livers as compared to the tumors

Oncogene Differential screening of hepatoblastoma cDNA libraries S Yamada et al 5907

Figure 4 Kaplan–Meier survival curves (n ¼ 59) in relation to the expression levels of PLK1 (median cutoff). The arbitrary median cutoff value was set as 13. The patients with high expression of PLK1 represented significantly poor prognosis than those with its low expression

Table 4 Univariate Cox regression analysis using PLK1(log) and dichotomous factors of b-catenin mutation, age, stage, and histology (n ¼ 59) Factor nP-value HR (95% CI)

PLK1(log) 59 0.015 1.62 (1.10, 2.40) b-catenin (mutant vs wild type) 58 0.27 1.85 (0.62, .5.56) Age (41vsr1 year) 55 0.76 1.22 (0.33, 4.52) Stage (3, 4vs 1, 2) 56 0.083 3.81 (0.84,17.2) Histology (poorly vs well) 53 0.025 4.48 (1.21, 16.6)

All variables with two categories, except PLK1(log); HR ¼ hazard ratio shows the relative of death of first category relative to second; CI ¼ confidence interval

or histology, but marginally related to survival after controlling both histology and stage (Table 5).

Discussion

HBL is one of the embryonal tumors in close relation to the normal as well as abnormal tissue development. To understand the molecular basis of the genesis of HBL, here we randomly cloned a large number of genes expressed in HBLs with or without AFP production and in a non-tumorous infant’s liver. Extensive screening for the differentially expressed genes between the tumors and their corresponding normal livers has successfully identified at least 86 genes including 40 with unknown function, which may potentially contribute to develop new therapeutic strategies against HBLs with poor prognosis. Figure 3 mRNA expression of PLK1 in HBLs and the corresponding normal livers measured by quantitative real-time RT–PCR. (a) The levels of PLK1 mRNA expression in HBLs and HBL cDNA libraries normal livers. The expression levels of PLK1 were determined by quantitative real-time RT–PCR analysis using 74HBL tissues and We have identified the genes with unknown function in 29 normal livers (see Materials and methods). The PLK1 approximately 8% of the total 10 431 clones obtained expression values were normalized by GAPDH. Open and closed from our oligo-capping cDNA libraries. The compar- circles represent alive and dead, respectively. Since the values of the PLK1 expression were skewed, a log transformation was used for ison of the frequently appeared genes in each libraries the expression values. The bars show mean values. (b) Correlation shows that expression profile is relatively similar of PLK1 expression between HBL and its corresponding normal between AFP-positive HBL and the normal part of the liver in 24paired samples infant’s liver, whereas it is very different between AFP- positive and AFP-negative tumors, in which many genes

Oncogene Differential screening of hepatoblastoma cDNA libraries S Yamada et al 5908 Table 5 Multivariable Cox regression analysis using PLK1(log) and dichotomous factors of b-catenin mutation, age, stage, and histology (n ¼ 50) Variable P-value Variable P-value Variable P-value

PLK1(log) 0.009 b-catenin (mutant vs. wild type) 0.51 PLK1(log) 0.005 Age (41vsr1 year) 0.92 PLK1(log) 0.019 Stage (3, 4vs 1, 2) 0.46 PLK1(log) 0.027 Histology (poorly vs well) 0.12 PLK1(log) 0.052 Histology (poorly vs well) 0.12 Stage (3, 4vs 1, 2) 0.47

All variables with two categories, except PLK1(log)

are downregulated (Table 2). In the library of the latter sponding normal livers. Surprisingly, 75 out of 86 genes tumor, vimentin, RNA-binding motif protein, Wnt are preferentially expressed in the latter tissues, and only inhibitory factor-1, dickkopf, and RAP1B are frequently 11 including RAN, PLK1, NPC1, and OLR1 known appeared, whereas they are hardly appeared in the other genes are expressed at high levels in HBLs. One of the libraries. Wissmann et al. (2003) have recently reported reasons of this result may be that many gene products, that WIF1 is downregulated in various cancers (prostate which are necessary for full function in the matured liver cancer, breast cancer, non-small-cell lung cancer, and metabolism, are dispensable for the malignant growth of bladder cancer), and suggested that loss of WIF1 the tumor except for the very limited genes. The results expression may be an early event in tumorigenesis in of some differentially expressed genes are consistent those tissues. It is notable that, in contrast to AFP- with those in the previous reports. von Horn et al. (2001) positive HBLs, the patient’s outcome of the tumor with have shown that the mRNA levels of insulin-like growth negative AFP is very poor, though the incidence of the factor-binding proteins including IGFBP-3 are decreased latter tumor is low (von Schweinitz et al., 1995). This in HBLs than in normal livers. Kinoshita and Miyata suggests that AFP-positive and AFP-negative HBLs (2002) have also reported that aldolase B mRNA is have a different genetic as well as biological back- downregulated in over 50% of 20 HCCs examined. They ground. In addition, recent reports have demonstrated proposed that the measurement of aldolase activity in that frequent mutation of the b-catenin gene and nuclear serum is useful to determine the number of collapsed accumulation of its protein product are one of the main hepatic cells in cirrhosis. Recently, evidences suggest causes of the tumorigenesis of HBL. The APC and Axin that not only mutant b-catenin but also wild-type b- genes are also mutated in some HBLs (Oda et al., 1996; catenin localize in the cellular nuclei of HBL as well as Miao et al., 2003; Thomas et al., 2003), indicating that some other cancers (Rimm et al., 1999; Takayasu et al., Wnt signaling pathway plays an important role in 2001). The increased expression of the Ran gene in causing the tumors, most of which are AFP-positive. HBLs might be correlated with the shuttling of b- Therefore, the poor-prognostic HBL without producing catenin and/or other related proteins between cytoplasm AFP might be caused by the particular mechanism and nucleus in the tumor cells. additional to or other than the abnormality of Wnt Owing to constitutive activation of Wnt signaling in signaling pathway. Although the appearance frequency most of the HBLs, the 86 genes differentially expressed of the genes in each library does not always reflect the between the tumor and its corresponding normal liver actual expression levels of each gene, it may at least in were expected to contain downstream target genes of part show the differences among the tumor subsets with Wnt signaling pathway that might regulate early stage of different genetic abnormalities. As our libraries contain the hepatic development. In this study, however, only many genes involved in liver development, normal liver the lymphocyte alpha-kinase (LAK) gene was found to be functions, and carcinogenesis, they must be useful for differentially expressed at high levels in HBLs with wild- making a liver-proper cDNA microarray to analyse type b-catenin and at low levels in those with b-catenin expression profiles of HBL, viral infection-induced mutation. LAK is a new class of protein kinases with a hepatitis, liver cirrhosis, and HCC. novel catalytic domain, but its precise function is currently unknown (Ryazanov et al., 1999). Thus, our Differentially expressed genes between HBLs and the result may suggest that the target genes of the Wnt corresponding normal livers signaling pathway are commonly affected in HBLs, regardless of the presence or absence of b-catenin cDNA microarray, which is often applied to a compre- mutation. hensive gene expression analysis, is able to detect many genes that are differentially expressed between tumors PLK1 as a prognostic indicator of HBL and normal tissues (Okabe et al., 2001; Nagata et al., 2003). However, it is expensive and needs further PLK1 (polo-like kinase 1), the human counterpart of confirmation of the selected genes by a semi-quantitative polo in Drosophila melanogaster and of CDC5 in RT–PCR or a real-time RT–PCR method. Therefore, Saccharomyces cerevisiae, encodes a serine/threonine using semi-quantitative RT–PCR and the specific kinase with polo-box domains (Clay et al., 1993). PLK1 primers of 1188 cDNAs, we have identified 86 genes is crucial for various events of mitotic progression differentially expressed between HBLs and their corre- including centrosome maturation (Lane and Nigg,

Oncogene Differential screening of hepatoblastoma cDNA libraries S Yamada et al 5909 1996), spindle function (Glover et al., 1996), activation (Uotani et al., 1998). A total of 74HBL samples (seven were of cyclin B/Cdc2 (Qian et al., 1998; Toyoshima- classified as being stage 1, 17 as stage 2, 26 as stage 3, 15 as Morimoto et al., 2001), and regulation of anaphase- stage 4, and nine were unknown stages) were used in this study. promoting complex (Kotani et al., 1998; Nigg, 1998). The tumors were staged according to the Japanese histopatho- Elevated expression of PLK1 is also found in different logical classification of HBL (Hata, 1990). From 1991 to 1999, HBLs had been treated by combination chemotherapy using types of adult cancers including non-small-cell lung cisplatin and THP-adriamycin according to the JPLT-1 cancer, head and neck tumors, esophageal carcinomas, protocol (Sasaki et al., 2002). After 2000, a more intensive melanomas, and colorectal cancers (Wolf et al., 1997; chemotherapeutic regimen, ITEC (ifosfamide, THP-adriamy- Knecht et al., 1999; Tokumitsu et al., 1999; Dietzmann cin, etoposide, and carboplatin), has been utilized for tumors et al., 2001; Takai et al., 2001), implying its critical role that prove resistant to the combination chemotherapy in the in tumorigenesis. In the present study, we have found JPLT-2 study. Among the 74tumor samples we examined, 36 that PLK1 is upregulated in primary HBLs, and that its and 35 tumor tissues were obtained prior to and after mRNA expression levels are significantly correlated chemotherapy, respectively, and the remaining three were with poor outcome of the patients. Multivariate Cox unknown. In the same sample set, 59 tumors were accom- regression analysis indicated that PLK1 expression panied by outcome information and used for making survival curves, among which 31 and 28 tissues were obtained prior to could be an independent prognostic factor from and after chemotherapy, respectively. Tumor histology was b-catenin mutation, age, stage, or histology. However, also classified according to Hata (1990). ‘Poor histology’ clinical stage did not show a significant correlation with indicates ‘poorly differentiated (embryonal type)’, and ‘well PLK1 expression, though it is one of the critical histology’ indicates ‘well-differentiated (fetal type)’. The prognostic markers. One of the possible reasons may informed consents were obtained in each institution or be that the 59 tumors we used for statistical analysis hospital. High molecular weight DNA and total RNA of include two unusual patients, one had stage 4tumor these samples were prepared as described previously (Ichimiya with good prognosis and another case had stage 1 tumor et al., 1999). with poor prognosis. These might have reduced the significance of the tumor stage in patients’ survival in Construction of oligo-capping cDNA libraries our sample set. Four oligo-capping cDNA libraries, two (HMFT, HYST) It is notable that, among the 1188 genes we have derived from HBLs with secretion of AFP, one (HKMT) from screened for differential expression, PLK1 is the only HBL without AFP secretion, and one (HMFN) from the one known oncogene overexpressed in the HBL tissues. corresponding normal liver, were constructed according to the Smith et al. (1997) have reported that constitutive method previously described (Suzuki et al., 1997). These were expression of PLK1 in NIH3T3 cells causes oncogenic approved by the institutional review board. The oligo-capping focus formation and forms tumors in nude mice. method enables full-length cDNA cloning with high efficiency. Furthermore, Liu and Erikson (2003) have recently The 12 000 cDNA clones in total were randomly picked up and shown that the application of small interfering RNA single-run sequencing was performed. Nucleotide sequence of which specifically depletes PLK1 expression in cancer both ends for each cDNA clone was homology-searched against the public nucleotide database using the BLAST cells inhibits cell proliferation, arrests cell cycle, and program at the National Center for Biotechnology Informa- induces apoptosis. Thus, PLK1 may play a crucial role tion (NCBI) (Genbank release 122, January 2001). in causing HBL and other cancers. It may be interesting to examine whether PLK1 is a target of b-catenin transported from the cytosol into the nucleus. The Differential screening of the genes by semi-quantitative disruption of PLK1 function could be a future RT–PCR therapeutic tool for the aggressive type of hepatoblas- The eight samples were selected as PCR templates to screen for tomas. the differentially expressed genes. Cases 58 and 81 were defined In conclusion, our HBL cDNA project has provided a as stage 2 HBL, cases 10, 67, 78, and 85 were in stage 3, case 14 large number of genes related to liver development, was in stage 4. Among those eight tumors, four (cases 14, 67, metabolism, and carcinogenesis. We are currently 78, and 81) had the mutant b-catenin, and the others (cases 10, 58, 77, and 85) not. Mutation analysis for b-catenin was applying these genes to the cDNA microarray system. performed as described previously (Takayasu et al., 2001). The Our cDNA resource should be an important tool to differential expression of the genes between the HBL and understand the molecular mechanism of the genesis of normal livers was confirmed at least twice using semi- HBL as well as to develop new diagnostic and quantitative RT–PCR. The individual gene-specific PCR therapeutic strategies against the aggressive tumors in primer sequences were determined by using Primer3 program the future. (provided at Washington University). For cDNA templates, 5 mg of total RNA was converted to cDNA using random primers (Takara, Otsu, Japan) with SuperScript II RNaseH- reverse transcriptase (Gibco BRL, Rockville, MD, USA). Materials and methods Those cDNAs were at first amplified with GAPDH primers for 27 cycles and the amounts of the PCR products were measured Clinical materials by ALF Expresst sequencer and normalized. The amplifica- Tumor tissues and their corresponding normal liver tissues tion was performed 35 or 40 cycles of 951C for 30 s, 57 or 59 or were frozen at the time of surgery and stored at À801C until 611C for 15 s and 721C for 60 s, and the final extension was at use. All specimens were provided from the Tissue Bank of the 721C for 5 min, using a Perkin-Elmer Thermalcycler 9700 Japanese Study Group for Pediatric Liver Tumor (JPLT) (Perkin-Elmer, Foster City, CA, USA). The PCR products

Oncogene Differential screening of hepatoblastoma cDNA libraries S Yamada et al 5910 were run on 2% agarose gels and stained with ethidium (Perkin-Elmer/Applied Biosystems). In all, 2 ml of cDNA was bromide. We defined the gene as differentially expressed when amplified in a final volume of 25 ml containing 1 Â Taqman it exhibits differential expression between the tumor and its PCR reaction buffer, 200 mM each dNTP, 0.9 mM each primer, corresponding normal liver in more than four out of the eight and 200 nM Taqman probe. The optional thermal cycling samples. condition was as follows: 40 cycles of a two-step PCR (951C for 15 s, 601C for 60 s) after the initial denaturation (951C for Northern blot analysis 10 min). Experiments were carried out in triplicate for each data point. In all, 25 mg of total RNA from the primary HBLs, HCC, and their corresponding normal livers were subjected to Northern analysis. Total RNA was prepared according to the method of Statistical analysis Chomczynski and Sacchi (1987). Total RNA was fractionated by electrophoresis on 1% agarose gel containing formalde- Statistical analyses were performed using Mann–Whitney’s U- hyde, transferred onto a nylon membrane filter, and immobi- test and Cox regression. A P-value of less than 0.05 was lized by UV crosslinking. The hybridization cDNA probe was considered significant. a 976-base pair human PLK1 cDNA fragment and labeled with [a-32P]-dCTP using the BcaBEST random priming kit (Takara Biomedicals). The filter was hybridized at 651Cina Acknowledgements solution containing 1 M NaCl, 1% SDS, 7.5% dextran sulfate, We are grateful to Shigeru Sakiyama and Toshinori Ozaki for 100 mg/ml of heat-denatured salmon sperm DNA, and radio- critical reading of the manuscript, and Yoko Nakamura and labeled probe DNA. Aiko Morohashi for experimental support. We thank Eriko Isogai, Naoko Sugimitsu, and Yuki Nakamura for preparing RNA and sequencing analysis, and Natsue Kitabayashi, Quantitative real-time RT–PCR of PLK1 Emiko Kojima, Emi Goto, and Hisae Murakami for technical The primer set for amplification of the PLK1 and probe assistance. We also thank the hospitals and institutions sequence are as follows: forward primer, 50-GCTGCACAAG collaborating with the Japanese Study Group for Pediatric AGGAGGAAA-30; reverse primer, 50-AGCTTGAGGTCTC- Liver Tumor (JPLT) for providing surgical specimens. This GATGAATAAC-30; probe, 50-CCTGACTGAGCCTGAGG work was supported in part by the fund from Hisamitsu CCCGATAC-TA-30. Taqman GAPDH control reagents (Per- Pharmaceutical Company and a grant-in-aid for Scientific kin-Elmer/Applied Biosystems) were used for the amplification Research on Priority Areas (C) ‘Medical Genome Science’ of GAPDH as recommended by the manufacturer. PCR was from the Ministry of Education, Culture, Sports, Science, and performed using ABI Prism 7700 Sequence Detection System Technology of Japan.

References

Albrecht S, Von Schweinitz D, Waha A, Kraus JA, Von Hata Y. (1990). Jpn. J. Surg., 20, 498–502. Deimling A and Pietsch T. (1994). Cancer Res., 54, 5041– Hata Y, Ishizu H, Ohmori K, Hamada H, Sasaki F, Uchino J, 5044. Inoue K, Naitoh H, Fujita M, Kobayashi T and Yokoyama Brady RO, Filling-Katz MR, Barton NW and Pentchev PG. S. (1991). Cancer, 68, 2566–2570. (1989). Neurol. Clin., 7, 75–88. Hsieh JC, Kodjabachian L, Rebbert ML, Rattner A, Small- Buendia MA. (1992). Adv. Cancer Res., 59, 167–226. wood PM, Samos CH, Nusse R, Dawid IB and Nathans J. Buendia MA. (2002). Med. Pediatr. Oncol., 39, 530–535. (1999). Nature, 398, 431–436. Chomczynski P and Sacchi N. (1987). Anal. Biochem., 162, Ichimiya S, Nimura Y, Kageyama H, Takada N, Sunahara M, 156–159. Shishikura T, Nakamura Y, Sakiyama S, Seki N, Ohira M, Clay FJ, McEwen SJ, Bertoncello I, Wilks AF and Dunn AR. Kaneko Y, McKeon F, Caput D and Nakagawara A. (1993). Proc. Natl. Acad. Sci. USA, 90, 4882–4886. (1999). Oncogene, 18, 1061–1066. Dietzmann K, Kirches E, Von Bossanyi, Jachau K and Idilman R, De Maria N, Colantoni A and Van Thiel DH. Mawrin C. (2001). J. Neurooncol., 53, 1–11. (1998). J. Viral. Hepat., 5, 285–299. Exelby PR, Filler RM and Grosfeld JL. (1975). J. Pediatr. Kinoshita M and Miyata M. (2002). Hepatology, 36, Surg., 10, 329–337. 433–438. Fukuzawa R, Umezawa A, Ochi K, Urano F, Ikeda H and Kinzler KW and Vogelstein B. (1996). Cell, 87, 159–170. Hata J. (1999). Int. J. Cancer, 82, 490–497. Knecht R, Elez R, Oechler M, Solbach C, Von Ilberg C and Giardiello FM, Petersen GM, Brensinger JD, Luce MC, Strebhardt K. (1999). Cancer Res., 59, 2794–2797. Cayouette MC, Bacon J, Booker SV and Hamilton SR. Koch A, Denkhaus D, Albrecht S, Leuschner I, Von (1996). Gut, 39, 867–869. Schweinitz D and Pietsch T. (1999). Cancer Res., 59, 269– Glover DM, Ohkura H and Tavares A. (1996). J. Cell Biol., 273. 135, 1681–1684. Kotani S, Tugendreich S, Fujii M, Jorgensen PM, Watanabe Goldstein JL and Brown MS. (1990). Nature, 343, 425–430. N, Hoog C, Hieter P and Todokoro K. (1998). Mol. Cell, 1, Gray SG, Eriksson T, Ekstrom C, Holm S, Von Schweinitz D, 371–380. Kogner P, Sandstedt B, Pietsch T and Ekstrom TJ. (2000). Lane HA and Nigg EA. (1996). J. Cell Biol., 135, 1701–1713. Br. J. Cancer, 82, 1561–1567. Li FP, Wendy AT, Seddon J and Holmes GE. (1987). J. Am. Haas JE, Muczynski KA, Krailo M, Ablin A, Land V, Vietti Med. Assoc., 257, 2475–2477. TJ and Hammond GD. (1989). Cancer, 64, 1082–1095. Li X, Adam G, Cui H, Sandstedt B, Ohlsson R and Ekstrom Hamanaka R, Kohno K, Seguchi T, Okamura K, Morimoto TJ. (1995). Oncogene, 11, 221–229. A, Ono M, Ogata J and Kuwano M. (1992). J. Biol. Chem., Liu X and Erikson RL. (2003). Proc. Natl. Acad. Sci. USA, 267, 13160–13165. 100, 5789–5794.

Oncogene Differential screening of hepatoblastoma cDNA libraries S Yamada et al 5911 Loftus SK, Morris JA, Carstea ED, Gu JZ, Cummings C, Suzuki Y, Yoshitomo-Nakagawa K, Maruyama K, Suyama A Brown A, Ellison J, Ohno K, Rosenfeld A, Tagle DA, and Sugano S. (1997). Gene, 200, 149–156. Pentchev PG and Pavan WJ. (1997). Science, 277, 232–235. Takai N, Miyazaki T, Fujisawa K, Nasu K, Hamanaka R and Mann JR, Lakin GE, Leonard JC, Rawlinson HA, Richard- Miyakawa I. (2001). Cancer Lett., 164, 41–49. son SG, Corkery JJ, Cameron AH and Shah KJ. (1978). Takayasu H, Horie H, Hiyama E, Matsunaga T, Hayashi Y, Arch. Dis. Child, 53, 366–374. Watanabe Y, Suita S, Kaneko M, Sasaki F, Hashizume K, Miao J, Kusafuka T, Udatsu Y and Okada A. (2003). Hepatol. Ozaki T, Furuuchi K, Tada M, Ohnuma N and Nakagawara Res., 25, 174–179. A. (2001). Clin. Cancer Res., 7, 901–908. Montagna M, Menin C, Chieco-Bianchi L and D’Andrea E. Taniguchi K, Roberts LR, Aderca IN, Dong X, Qian C, (1994). J. Cancer Res. Clin. Oncol., 120, 732–736. Murphy LM, Nagorney DM, Burgart LJ, Roche PC, Smith Morin PJ, Sparks AB, Korinek V, Barker N, Clevers H, DI, Ross JA and Liu W. (2002). Oncogene, 21, 4863–4871. Vogelstein B and Kinzler KW. (1997). Science, 275, 1787– Thomas D, Pritchard J, Davidson R, McKiernan P, Grundy 1790. RG and de Ville de Goyet J. (2003). Eur. J. Cancer, 39, Nagata T, Takahashi Y, Ishii Y, Asai S, Nishida Y, Murata A, 2200–2204. Koshinaga T, Fukuzawa M, Hamazaki M, Asami K, Ito E, Tokumitsu Y, Mori M, Tanaka S, Akazawa K, Nakano S and Ikeda H, Takamatsu H, Koike K, Kikuta A, Kuroiwa M, Niho Y. (1999). Int. J. Oncol., 15, 687–692. Watanabe A, Kosaka Y, Fujita H, Miyake M and Toyoshima-Morimoto F, Taniguchi E, Shinya N, Iwamatsu A Mugishima H. (2003). Cancer Genet. Cytogenet., 145, 152– and Nishida E. (2001). Nature, 410, 215–220. 160. Uotani H, Yamashita Y, Masuko Y, Shimoda M, Murakami Nigg EA. (1998). Curr. Opin. Cell Biol., 10, 776–783. A, Sakamoto T, Tazawa K and Tsukada K. (1998). J. Oda H, Nakatsuru Y, Imai Y, Sugimura H and Ishikawa T. Pediatr. Surg., 33, 639–641. (1995). Int. J. Cancer, 60, 786–790. Van Tornout JM, Buckley JD, Quinn JJ, Feusner JH, Krailo Oda H, Imai Y, Nakatsuru Y, Hata J and Ishikawa T. (1996). MD, King DR, Hammond GD and Ortega JA. (1997). J. Cancer Res., 56, 3320–3323. Clin. Oncol., 15, 1190–1197. Okabe H, Satoh S, Kato T, Kitahara O, Yanagawa R, Velculescu VE, Vogelstein B and Kinzler KW. (2000). Trends Yamaoka Y, Tsunoda T, Furukawa Y and Nakamura Y. Genet., 16, 423–425. (2001). Cancer Res., 61, 2129–2137. Von Horn H, Tally M, Hall K, Eriksson T, Ekstrom TJ and Okubo K, Hori N, Matoba R, Niiyama T, Fukushima A, Gray SG. (2001). Cancer Lett., 162, 253–260. Kojima Y and Matsubara K. (1992). Nat. Genet., 2, 173– Von Schweinitz D, Hecker H, Harms D, Bode U, Weinel P, 179. Burger D, Erttmann R and Mildenberger H. (1995). J. Ortega JA, Krailo MD, Haas JE, King DR, Ablin AR, Quinn Pediatr. Surg., 30, 845–852. JJ, Feusner J, Campbell JR, Lloyd DA, Cherlow J and Von Schweinitz D, Wischmeyer P, Leuschner I, Schmidt D, Hammond GD. (1991). J. Clin. Oncol., 9, 2167–2176. Wittekind C, Harms D and Mildenberger H. (1994). Eur. J. Park WS, Oh RR, Park JY, Kim PJ, Shin MS, Lee JH, Kim Cancer, 30A, 1052–1058. HS, Lee SH, Kim SY, Park YG, An WG, Kim HS, Jang JJ, Wang J, Shou J and Chen X. (2000). Oncogene, 19, 1843–1848. Yoo NJ and Lee JY. (2001). J. Pathol., 193, 483–490. Weber RG, Pietsch T, Von Schweinitz D and Lichter P. (2000). Polakis P. (1999). Curr. Opin. Genet. Dev., 9, 15–21. Am. J. Pathol., 157, 571–578. Qian YW, Erikson E, Li C and Maller JL. (1998). Mol. Cell. Wei Y, Fabre M, Branchereau S, Gauthier F, Perilongo G and Biol., 18, 4262–4271. Buendia MA. (2000). Oncogene, 19, 498–504. Rainier S, Dobry CJ and Feinberg AP. (1995). Cancer Res., 55, Weinberg AG and Finegold M. (1983). Hum. Pathol., 14, 512– 1836–1838. 537. Ribbeck K, Lipowsky G, Kent HM, Stewart M and Gorlich Wissmann C, Wild PJ, Kaiser S, Roepcke S, Stoehr R, D. (1998). EMBO J., 17, 6587–6598. Woenckhaus M, Kristiansen G, Hsieh JC, Hartmann A, Rimm DL, Caca K, Hu G, Harrison FB and Fearon ER. Knuechel R, Rosenthal A and Pilarsky C. (2003). J. Pathol., (1999). Am. J. Pathol., 154, 325–329. 201, 204–212. Ryazanov AG, Pavur KS and Dorovkov MV. (1999). Curr. Wolf G, Elez R, Doermer A, Holtrich U, Ackermann H, Biol., 9, R43–45. Stutte HJ, Altmannsberger HM, Rubsamen-Waigmann H Sasaki F, Matsunaga T, Iwafuchi M, Hayashi Y, Ohkawa H, and Strebhardt K. (1997). Oncogene, 14, 543–549. Ohira M, Okamatsu T, Sugito T, Tsuchida Y, Toyosaka A, Xu XR, Huang J, Xu ZG, Qian BZ, Zhu ZD, Yan Q, Cai T, Nagahara N, Nishihira H, Hata Y, Uchino J, Misugi K and Zhang X, Xiao HS, Qu J, Liu F, Huang QH, Cheng ZH, Li Ohnuma N. (2002). J Pediatr Surg., 37, 851–856. NG, Du JJ, Hu W, Shen KT, Lu G, Fu G, Zhong M, Xu Smith MR, Wilson ML, Hamanaka R, Chase D, Kung H, SH, Gu WY, Huang W, Zhao XT, Hu GX, Gu JR, Chen Z Longo DL and Ferris DK. (1997). Biochem. Biophys. Res. and Han ZG. (2001). Proc. Natl. Acad. Sci. USA, 98, 15089– Commun., 234, 397–405. 15094. Sudhof TC, Russell DW, Brown MS and Goldstein JL. (1987). Yun K, Jinno Y, Sohda T, Niikawa N and Ikeda T. (1998). J. Cell, 48, 1061–1069. Pathol., 185, 91–98.

Oncogene